Jérôme Dockès

Thanks for this nice summary, @kchawla-pi! I think it would be better if we reached a decision (at least in principle) before you do the actual work of writing the...

also could you list the considered hooks and what they do? the discussion mentions code quality and linting so this is not just about formatting code with black any more?

btw the discussion linked in this issue involved bugs, a function's signature, and the quality of unit tests so the PR's delay was not due to formatting but more fundamental...

> To start with: thanks! In the link I also see pylint; we may want to add that to the list of candidates, I find it often has helpful warnings

I just meant we shouldn't overestimate the importance of formatting. But you are right, we do strive to enforce pep8. although note that black cannot enforce some of the most...

> Just masking values below 4 produces incorrect representation of the data (second plot). The figure only shows values that almost exceed twice the threshold value (compared to original in...

is it indispensable for some example in the gallery? otherwise I would vote for not adding new fetchers https://nilearn.github.io/stable/development.html#how-to-contribute-a-dataset-fetcher

"Downloaders for new atlases are welcome if they comes with an example." you are right! downloads and new examples put strain on circle ci and maintenance so we may want...

> The original code will project the data back to the space of the maps. The default resampling strategy is data, and this will create a lot of confusion. to...

> We've discussed this elsewhere (perhaps in a dev call?), but I can't find any notes on it maybe you are referring to the discussion that started [here](https://github.com/nilearn/nilearn/pull/2722#pullrequestreview-608823691) ? I...